Skip to content

[ROCm][Kernel] W4A16 prefill: optimize dequant#985

Open
mgehre-amd wants to merge 6 commits into
gfx11from
matthias.triton-w4a16-skinny-packedsb
Open

[ROCm][Kernel] W4A16 prefill: optimize dequant#985
mgehre-amd wants to merge 6 commits into
gfx11from
matthias.triton-w4a16-skinny-packedsb

[ROCm][Kernel] W4A16: bench real weight layout; re-tune fp16 tiles

eba17b1
Select commit
Loading
Failed to load commit list.